Gambler's Ruin and the Mathematics of Runway

The gambler’s ruin problem is two centuries old and still the best short argument against confusing expected value with survival. The setup: you start with a bankroll of $i$ units and bet one unit at a time, winning with probability $p$ and losing with probability $q = 1 - p$. You stop when you reach a target of $N$ units — or when you hit zero and are ruined.

The ruin probabilities

Let $P_i$ be the probability of reaching $N$ before ruin, starting from $i$. Conditioning on the first bet gives a linear recurrence:

$$P_i = p \, P_{i+1} + q \, P_{i-1}, \qquad P_0 = 0, \quad P_N = 1.$$

Solving it (the characteristic root is $r = q/p$) yields the closed form:

$$P_i = \begin{cases} \dfrac{1 - (q/p)^i}{1 - (q/p)^N} & \text{if } p \neq q, \\[2ex] \dfrac{i}{N} & \text{if } p = q = \tfrac{1}{2}. \end{cases}$$

Two things in this formula deserve to be stared at.

The fair game is not safe

Set $p = 1/2$. Your chance of doubling a bankroll of $i = 10$ to $N = 20$ is exactly $10/20 = 1/2$ — fair enough. But let the target grow, $N \to \infty$, and $P_i = i/N \to 0$. A fair game played indefinitely against a much richer opponent ends in ruin with probability one. The casino does not need an edge to beat you; it only needs more chips and more patience.

A small edge compounds — in both directions

Now set $p = 0.49$, barely unfavorable. With $i = 10$ and $N = 20$,

$$P_{10} = \frac{1 - (0.51/0.49)^{10}}{1 - (0.51/0.49)^{20}} \approx 0.40,$$

and the miss compounds brutally as the horizon stretches: the ruin probability from a finite bankroll against an infinite opponent is $(p/q)^i$ when $p < q$… which tends to certainty fast. Symmetrically, a favorable game ($p > 1/2$) leaves ruin probability $\left(q/p\right)^i$ — exponentially small in the bankroll $i$, but never zero. Edge does not protect you; capitalization protects you.

Runway is a ruin problem

Replace “bankroll” with months of runway, “bet” with a month of operations, and the theorem reads differently: a venture with positive expected value and thin reserves is still overwhelmingly likely to die if variance is high and the horizon is long. The same holds for personal energy — a sustainable pace with reserves beats a brilliant sprint at $i = 1$.

The Stoic angle, because this blog owes you one: Seneca’s praemeditatio malorum is bankroll management for the psyche. You rehearse losses in advance not to be gloomy, but so that no single draw — however bad — takes your $i$ to zero.